Original title: Doppio: A Dataset for Contactless Weight Estimation of Falling Particles
Simon Kiefhaber, Jan-Martin O. Steitz, Julia Grabinski, Christoph Reich +4 · September 2026
Measuring the weight of falling powder in industrial settings is difficult without touching it. Regular scales work well only when objects stay still, and current touch-free tools are often expensive, complicated, and limited to single uses. To solve this, researchers tested whether computer vision could estimate mass just by looking. They created a new video dataset called Doppio, focusing on the real-world example of falling ground coffee. Each video frame comes paired with exact weight measurements. The team tested several deep learningA subset of artificial intelligence where algorithms learn patterns from large datasets to make predictions. systems, which are smart algorithms trained to recognize patterns. Some analyzed single images, while others tracked motion over time. They found that these vision models can accurately estimate the total accumulated weight of falling particles.
Imagine you're trying to estimate how much sugar is pouring out of a jar just by watching it fall into a bowl. If you can't use a scale, you have to rely on visual cues like the stream's thickness and the pile's shape. Here, the researchers did something similar with coffee powder: they recorded videos and used computer programs to guess the weight from the footage, turning visual information into accurate measurements.
Factories that process powders need to measure mass without touching the material to avoid contamination and save time. Current methods are expensive or only work for specific situations. This research shows that standard cameras and software could offer a cheaper, more flexible solution.
The abstract indicates this is a case study focused on coffee grinding, so the approach might not generalize to all particles or settings without further testing. It establishes a foundation but does not address all real-world deployment challenges.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Doppio: A Dataset for Contactless Weight Estimation of Falling Particles — Simon Kiefhaber, Jan-Martin O. Steitz, Julia Grabinski et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.